At Databricks, we often use the phrase “the future is open” to refer to technology; it reflects our belief that open data architecture will win out and subsume proprietary ones (we just set a new official record on TPC-DS). But "open" isn't just about code. It's about how we as an industry operate and foster debate. Today, many companies in tech have tried to control the narrative on their products' performance through a legal maneuver called the DeWitt Clause, which prevents comparative benchmarking. We think this practice is bad for customers and bad for innovation, and it's time for it to go. That’s why we are removing the DeWitt Clause from our service terms, and calling upon the rest of the industry to follow.
What is the DeWitt Clause?
According to Wikipedia, “the original DeWitt Clause was established by Oracle at the behest of Larry Ellison. Ellison was displeased with a benchmark study done by David DeWitt in 1982, …, which showed that Oracle’s system had poor performance.”
Professor David DeWitt is a well-known database researcher who pioneered many of the technologies modern-day database systems depend on. Early on in his career, DeWitt spent a lot of time benchmarking different commercial database systems to understand and push forward the state-of-the-art. He published papers that demonstrated the strengths and weaknesses of such systems, including Oracle.
Triggered by this research, Oracle created the DeWitt Clause, a new provision that prohibits people (researchers, scientists, or competitors) from publishing any benchmarks of Oracle’s database systems.
Over time, the DeWitt Clause became a standard feature of most database vendors’ license agreements. It’s a primary reason you often see benchmarks comparing anonymous systems, sometimes referred to as DBMS-X, in research papers and why many benchmarks are completely absent.
Almost 40 years have passed since the introduction of the original DeWitt Clause – 40 years of DeWitt's name, the name of a tech pioneer who was pro benchmarking, being synonymous with preventing its use.
An open era calls for open terms: introducing the “DeWitt Embrace Clause"
Like many vendors – from classic ones like Oracle to newer ones like Snowflake – we too included a DeWitt Clause in the past, as it was the industry standard.
But standard isn’t good enough. We owe our users more.
We believe our users should have access to transparent benchmarks to decide which products are best for them. We also don’t think we should get angry at benchmarks in which we do poorly, as long as they were conducted in a transparent, fair fashion. We view those as opportunities to improve our product.
As a result, we have removed the DeWitt clause from our service terms.
But getting rid of our DeWitt clause isn’t enough; if you want to be able to see real, competitive benchmarks, you need to demand that no one relies on a DeWitt clause to stifle competition.
That’s why we’re also introducing what we refer to as a “DeWitt Embrace Clause”: if a competitor or vendor benchmarks Databricks or instructs a third party to do so, this new provision invalidates the vendor’s own DeWitt Clause if there’s any, to allow us to benchmark them and explicitly name them in the benchmark. We are not alone in this. For example, Azure and AWS have adopted similar clauses.
Today we call on the rest of the industry to follow in DeWitt's footsteps.
We believe companies should win or lose because their products are better or worse. Because their engineers innovate. Not because their lawyers stagnate.
The future is open.